The Agent Skills Directory

[SAFE]: The skill consists entirely of natural language instructions and documentation for using established tools (recall, retain, reflect). It does not contain executable code, external scripts, or obfuscated content.
[DATA_EXPOSURE]: The documentation includes a security best practice, explicitly instructing the agent not to retain secrets (such as API keys or passwords) or anything the user has requested to keep private.
[INDIRECT_PROMPT_INJECTION]: While the skill establishes a memory system that could theoretically store and later recall malicious instructions provided by a user in a previous session, this is a known risk inherent to all memory-based AI systems. The skill itself does not provide dangerous capabilities that would make this injection particularly exploitable beyond standard agent behavior.

hindsight-memory